TomePatchModel - ComfyUI Built-in Node Documentation

Complete documentation for the TomePatchModel node in ComfyUI. Learn its inputs, outputs, parameters and usage.

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

The TomePatchModel node applies Token Merging (ToMe) to a diffusion model to reduce computational requirements during inference. It works by selectively merging similar tokens in the attention mechanism, allowing the model to process fewer tokens while maintaining image quality. This technique helps speed up generation without significant quality loss.

Inputs

Parameter	Data Type	Required	Range	Description
model	MODEL	Yes	-	The diffusion model to apply token merging to
ratio	FLOAT	No	0.0 - 1.0	The ratio of tokens to merge (default: 0.3)

Outputs

Output Name	Data Type	Description
model	MODEL	The modified model with token merging applied